U Kv f @s.ddlZddlZddlmZGdddZdS)N)statsc@s~eZdZdZdddZdddZeddZed d Zed d Z ed dZ eddZ dddZ dddZ dddZdS)PredictionResultsa Prediction results Parameters ---------- predicted_mean : {ndarray, Series, DataFrame} The predicted mean values var_pred_mean : {ndarray, Series, DataFrame} The variance of the predicted mean values dist : {None, "norm", "t", rv_frozen} The distribution to use when constructing prediction intervals. Default is normal. df : int, optional The degree of freedom parameter for the t. Not used if dist is None, "norm" or a callable. row_labels : {Sequence[Hashable], pd.Index} Row labels to use for the summary frame. If None, attempts to read the index of ``predicted_mean`` NcCst||_t||_||_||_|dkr:t|dd|_|jdk |_|dkr^|dk r^td|dksn|dkr~t j |_ d|_ n@|dkrt j |_ |jf|_ n$t|t jjr||_ d|_ ntddS)Nindextz$df must be None when dist is not "t"normz/dist must be a None, "norm", "t" or a callable.)npasarray_predicted_mean_var_pred_meanZ_df _row_labelsgetattr _use_pandas ValueErrorrrdist dist_argsr isinstance distributionsZ rv_frozen)selfpredicted_mean var_pred_meanrZdf row_labelsrrC/tmp/pip-unpacked-wheel-2v6byqio/statsmodels/tsa/base/prediction.py__init__s&    zPredictionResults.__init__cCs8|js |S|jdkr&tj||j|dStj||j|dS)N)rname)rcolumns)rndimpdSeriesr DataFrame)rvaluerrrrr _wrap_pandas:s  zPredictionResults._wrap_pandascCs|jS)z$The row labels used in pandas-types.)r rrrrrAszPredictionResults.row_labelscCs||jdS)zThe predicted meanr)r"r r#rrrrFsz PredictionResults.predicted_meancCs |jjdkr|jS||jdS)z"The variance of the predicted meanr)r rr"r#rrrrKs zPredictionResults.var_pred_meancCsN|jj}|dkrt|j}n$|dkr:t|jj}ntd||dS)z,The standard deviation of the predicted meanrzvar_pre_mean must be 1 or 3 dimmean_se)r rrsqrtTZdiagonalNotImplementedErrorr")rrvaluesrrrse_meanRszPredictionResults.se_meancCs"|j|j}t|tjrd|_|S)z9The ratio of the predicted mean to its standard deviationtvalues)rr+rrrr)rvalrrrr,^s  zPredictionResults.tvaluesr two-sidedcCs|j||j}|dkr8|jjt|f|jd}nD|dkrV|jj|f|j}n&|dkrt|jj|f|j}ntd||fS)a8 z- or t-test for hypothesis that mean is equal to value Parameters ---------- value : array_like value under the null hypothesis alternative : str 'two-sided', 'larger', 'smaller' Returns ------- stat : ndarray test statistic pvalue : ndarray p-value of the hypothesis test, the distribution is given by the attribute of the instance, specified in `__init__`. Default if not specified is the normal distribution. )r.z2-sidedZ2sr$)Zlargerl)Zsmallerszinvalid alternative) rr+rZsfrabsrZcdfr)rr! alternativestatZpvaluerrrt_testfs zPredictionResults.t_test皙?cCsh|j}|jjd|df|j}|j||}|j||}t||f}|jrd|j|ddgdS|S)a  Confidence interval construction for the predicted mean. This is currently only available for t and z tests. Parameters ---------- alpha : float, optional The significance level for the prediction interval. The default `alpha` = .05 returns a 95% confidence interval. Returns ------- pi : {ndarray, DataFrame} The array has the lower and the upper limit of the prediction interval in the columns. rg@lowerupper)r) r+rZppfrrrZ column_stackrr")ralphaseqr6r7cirrrconf_intszPredictionResults.conf_intcCsPt|j|d}|dddf|dddf}}|j|j||d}t|S)a Summary frame of mean, variance and confidence interval. Returns ------- DataFrame DataFrame containing four columns: * mean * mean_se * mean_ci_lower * mean_ci_upper Notes ----- Fixes alpha to 0.05 so that the confidence interval should have 95% coverage. )r8Nrr)Zmeanr&Z mean_ci_lowerZ mean_ci_upper)rr r<rr+rr )rr8Zci_meanr6r7Z to_includerrr summary_frames"zPredictionResults.summary_frame)NNN)NN)rr.)r5)r5)__name__ __module__ __qualname____doc__rr"propertyrrrr+r,r4r<r=rrrrrs&        ! r)ZnumpyrZpandasrZscipyrrrrrrs